Bill > SB474

This bill, the Protecting Louisiana's Infrastructure from Artificial Intelligence Risk Act, establishes new regulations for "large frontier developers," which are defined as companies with over $500 million in annual gross revenue that train advanced artificial intelligence (AI) models, referred to as "frontier models." These models are capable of generating outputs that can influence real-world environments and are trained using significant computing power. The Act mandates that these developers create and publish a "frontier AI framework," a set of protocols designed to identify, assess, and mitigate "catastrophic risks," which are defined as foreseeable and material risks that a frontier model could cause widespread death, serious injury, or over $1 billion in property damage through actions like facilitating chemical or biological weapon development or enabling cyberattacks on critical infrastructure, such as energy grids or healthcare systems. This framework must include specific measures for cybersecurity and biosecurity, involve third-party assessments, and be updated annually. Developers must also publish transparency reports before deploying new or significantly modified frontier models, detailing risk assessments and mitigation strategies. Furthermore, the bill requires the reporting of "critical safety incidents"—events like unauthorized access to model weights, loss of control of a frontier model, or specific assistance in creating weapons—to the Louisiana Department of Justice within strict timeframes, with immediate reporting required for imminent threats. The Act also includes whistleblower protections for employees who report potential dangers related to frontier models and establishes penalties for non-compliance, including civil fines of up to $10 million for repeated violations, while also exempting certain sensitive reports from public records laws until July 1, 2031.